When asked about using edge uniques vs mw user ID for the Temp Accounts WE 1.8.6 experiment (https://growthbook.wikimedia.org/experiment/exp_10ipnmqzch3jt#overview), @Michael shared:

We need to track the subject id across both the temporary account and the (converted) named account. However, these two accounts will have different user-ids and so with a logged-in experiment they would translate to two different subject-ids, which would be a problem.

So using edge-uniques is the approach to track the user across the boundary of two accounts (temp->named)

it would be great if we could raise the amount of enrolled users (that's why we picked NCS) to 50% due to this being an account creation experiment

AC

• Higher traffic allocation limit of 50% is approved
• we-1-8-tempuser-post-edit experiment is configured in TK UI with the same parameters as it is configured in GB: https://test-kitchen.wikimedia.org/experiment/we-1-8-tempuser-post-edit
  • Same contextual attributes selected
  • Same target wikis selected
  • Same identifier type (edge unique NCS) selected
• Manual (backend) overrides:
  • Start date is 2026-06-30 14:30 UTC (earlier today)
  • Traffic allocation is 50%
• Activate experiment ONLY if the above ACs have been satisfied:
  • Approval
  • Config
  • Manual override
• Once experiment is active in TK UI, switch experiment config providers from GB to TK UI by tagging the experiment (in GB) with external imported tag
• Verify that https://test-kitchen.wikimedia.org/api/v1/experiments?format=config&authority=varnish shows the experiment with bigger bucket ranges